While English is the most commonly used language for market research, surveys and customer feedback in India, it does not reach the subcontinent’s entire population. Therefore, many questionnaires are bilingual, offering the respondent a choice between English and an indigenous Indian language. This, however, presupposes that answers to items are not influenced by people’s language proficiencies and response styles in different languages. This paper aims to examine whether market researchers in India should be careful about nonrandom measurement error caused by language response bias.
English and Kannada questionnaires are administered in a test-retest scenario to 160 respondents in the Indian Tier-II city of Mysore. The data evaluation is organized by dispositional (language proficiency in English and Kannada) and situational influences (language of the questionnaire in English or Kannada); a series of tests to elucidate language response bias is conducted.
While the significance of the two-tailed English-Kannada paired-sample tests is borderline, a more detailed look reveals surprising differences for the dispositional as well as situational linguistic influences. Moreover, the response style peculiarities in the Indian multilingual environment are not always consistent with differences in other international bilingual environments.
High-quality data are central to all empirical research, but situational and dispositional language response bias seems to contaminate questionnaires in the Indian multi-lingual environment. This study highlights the effect and provides Indian market researchers with some first strategies for managing the challenge.
